The Neuroepidemiology Branch (NEB) and an international group of expert and nonexpert neurologists estimated the accuracy in the diagnosis of different parkinsonian and dementia disorders using clinical material confirmed by postmortem examination. These studies are relevant for the development of standardized criteria for the diagnosis of each of these disorders. Only well-defined populations lead to firm scientific conclusions in genetic, clinical drug trials and analytic epidemiologic clinical studies. The NEB used neural network methodology to investigate which histologic features have diagnostic relevance and whether alternative diagnostic classifications are needed. Identification of predictors for the diagnosis of corticobasal degeneration, progressive supranuclear palsy, Pick's disease, multiple system atrophy, dementia with well-defined populations for research studies are being examined.